Thesis work and research project

Deploying critical services (e.g. VPN etc.) over the Internet X-domain topology requires to be able to guarantee end-to-end QoS. For this, and to guarantee privacy, Service Level Agreements (SLAs), also called QoS contracts, are committed pairwise between domains. If local QoS control issues inside on each participating domain can be considered solved, open problems persist in the eld of end-to-end QoS provisioning and monitoring for multi-domain services. The key factors to be taken into account are the heterogeneity, independence and privacy requirements of the individual domains. Before establishing a service, a negotiation has to occur: it consists in selecting a chain of pair-wise commitments (Fig.1), that satis es the end-toend QoS requirements, given that global QoS is subject to cumulation e ects: for instance, delays on each domain sum up along a path. We use the term of QoS budget to re ect how the customer's tolerance (w.r.t to end-to-end delay, jitter, etc.) is consumed by the actual services on each component, leaving a reduced margin for the rest of the chain. Once a contract chain is selected by the negotiation, QoS contracts are reserved and the service is established. Each domain monitors the QoS contract agreed on with its neighbor in the path, and also its local QoS provisioning. If the domain observes repeated contract violations by its neighbor, it can activate a re-negotiation process to select another partner in the path and thus changing the sub-path of the contract chain. So, the problem addressed is to satisfy a QoS budget by a contract chain taking into account i) cumulative e ect of QoS parameters; ii) domain in-